Represent language, retrieve evidence, and build grounded systems that expose their sources.
Nearest-neighbor retrieval often uses cosine similarity to rank semantically related vectors.
A reusable technical reading framework for moving from vocabulary to working understanding. Use it alongside the formulas and papers below.
Represent language, retrieve evidence, and build grounded systems that expose their sources. Start by naming the representation, objective, and source of evidence in any system you study. This habit separates a useful model explanation from a list of buzzwords.
Use the learning outcomes as checks for understanding: Compare tokenization, embeddings, retrieval, and generation; Design a citation-aware RAG pipeline; Evaluate semantic similarity and factuality. For each claim, ask what assumptions make it true, what data it needs, and how it could fail.
Translate each equation into a small experiment before treating it as memorized knowledge. Embedding similarity: Nearest-neighbor retrieval often uses cosine similarity to rank semantically related vectors.
Read primary work with a repeatable lens: problem, method, data, measurement, limits, and what would change your mind. This section starts with “Efficient Estimation of Word Representations in Vector Space”, “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”, “GloVe: Global Vectors for Word Representation”. Follow citations outward only after you can explain the central claim in your own words.
Implement a minimal example using Embedding similarity; record inputs, outputs, and a failure case.
Define a baseline, a success metric, a stress test, and the evidence required before you trust an improvement.
Choose one paper from the library, reproduce a small claim, then write down the questions its evaluation leaves open.
Primary sources and peer-reviewed work to deepen this page.
T. Mikolov et al.
Introduces efficient word2vec objectives that learn useful distributional representations.
P. Lewis et al.
Combines a parametric generator with a non-parametric document index to ground answers in retrieved evidence.
J. Pennington · R. Socher · C. Manning
Learns word vectors from global co-occurrence statistics and provides strong semantic structure.